Adsorption of Polar Enantiomers in Achiral Zeolites - The Journal of

Dec 31, 2012 - Tom R. C. Van Assche , Tim Duerinck , Juan José Gutiérrez Sevillano , Sofia Calero , Gino V. Baron , and Joeri F. M. Denayer. The Jou...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/JPCC

Adsorption of Polar Enantiomers in Achiral Zeolites Ana Martin-Calvo,† Sofia Calero,† Johan. A. Martens,‡ and Titus S. van Erp*,§ †

Department of Physical, Chemical, and Natural Systems, University Pablo de Olavide, Ctra. de Utrera, km.1, 41013 Sevilla, Spain Centrum voor Oppervlaktechemie en Katalyse, KU Leuven, Kasteelpark Arenberg 23, B-3001 Leuven, Belgium § Department of Chemistry, Norwegian University of Science and Technology, 7941 Trondheim, Norway ‡

ABSTRACT: We used molecular simulation techniques to analyze the enantioselective adsorption of chiral polar molecules in achiral zeolites. As a representative chiral molecule, CHClFBr was used in this study. We analyzed the adsorption of racemic and scalemic mixtures of these molecules into different zeolites topologies such as MFI and MEL, both structures with intersecting channels, and FER and TON, frames with nonintersecting channels. In contrast to previous findings on apolar molecules, we found that cations are not essential for imposing heteroselective adsorption, and homoselective adsorption was never observed. We derived the mechanism behind the selectivity, which turns out to be fundamentally different from that observed for apolar adsorbates. Our model is able to explain why apparently similar zeolite topologies can still have strong differences in their adsorption behavior.



INTRODUCTION Molecular simulations have proven to be a useful tool for studying the chiral separation of enantiomeric mixtures. From the industrial point of view, this separation is extremely important because, in many cases, just one of the enantiomers has the desired effect.1−3 The use of zeolites as adsorbents has been considered because of their favorable characteristics. For this purpose, synthesis studies have focused on the creation of chiral zeolites,4−7 but most of them have been found to be thermally unstable. Because not many chiral zeolites are available, the recently reported enantioselectivity in achiral zeolites opens the door to propitious alternatives for separation processes using these zeolites. The approach seems to contradict the general idea that an adsorbent needs to be chiral itself to separate a chiral mixture. However, the essential symmetry-breaking aspect is provided here not by the adsorbent but by the adsorbate because it considers mixtures in which one of the enantiomers appears in a higher concentration than its mirror image (scalemic mixture). The simulation results on this effect have been improved thanks to a Monte Carlo (MC) scheme newly developed by van Erp et al.8 for studying chiral mixtures in nanoporous materials. Using this technique, Caremans et al.9 and van Erp et al.10 demonstrated that chiral amplification of scalemic mixtures is possible as a result of adsorption in achiral aluminumsubstituted zeolites. If one type of enantiomer is already present in a higher concentration than its counterpart in the gas phase, this relative concentration might either increase (homoselective), decrease (heteroselective), or remain the same (aselective) upon adsorption. The type of adsorption depends on the zeolite topology, the aluminum distribution, and the type of cations used to balance the net charge of the system. Technological applications of this effect will mainly lie in chiral © 2012 American Chemical Society

purification rather than separation, but in some circumstances, chiral separation of racemic mixtures is possible as well.11 The occurrence of enantioselective adsorption of chiral alkanes was explained by the chiral cell model.9,10 In this model, the molecules do not interact directly, but rather interact indirectly through the mobile cations that are interconnected through their long-range Coulombic force. The chiral cell model could rationalize the absence of selectivity and the occurrence of either homo- or heteroselective adsorption in all studied aluminosilicate frameworks. In this work, we aimed to study the effect of polarity. Chlorofluorobromomethane, CHClFBr, was our test molecule for this study. It is one of the simplest chiral organic molecules, and isomers of this molecule have previously been used to study adsorption in chiral zeolites.12 The achiral zeolite topologies that we considered are (a) MFI and MEL, structures characterized by longitudinal channels interconnected by zigzag channels, and (b) FER and TON, characterized by being formed only by longitudinal channels. In Figure 1, energy grids of the two types of topologies are depicted. In addition, we analyzed several aluminum-substituted forms of MFI (Al-1 Al-2, Al-3, Al-4, Al-5, Al-6, Al-8, and Al-11)9 using calcium as the nonframework cation. Note that these substitutions do not induce chirality into the zeolite crystal. It turns out that the enantioselective adsorption behavior of chiral polar molecules and the mechanisms behind it are very different from those of apolar molecules. Received: November 12, 2012 Revised: December 23, 2012 Published: December 31, 2012 1524

dx.doi.org/10.1021/jp3111937 | J. Phys. Chem. C 2013, 117, 1524−1530

The Journal of Physical Chemistry C

Article

To analyze the energetics of our simulations, we made some comparative studies in vacuo of minimized clusters. These minimizations were calculated using the Baker method with a root-mean-square (RMS) and maximum (MAX) gradient tolerance of 10−6.



RESULTS AND DISCUSSION The computed adsorption isotherms of enantiopure CHClFBr in the four structures under study are shown in Figure 2. MEL, MFI and FER zeolites exhibited the highest loadings at around 2 mol/kg, whereas TON zeolite reached saturation at approximately 1.2 mol/kg.

Figure 1. Energy grid representation of the different topologies considered in this work: (left) Intersecting channels (MFI and MEL) and (right) longitudinal channels (FER and TON). The interior of the energy surface (gray) delimits the pore volume where molecules can be adsorbed. The oxygen and silicon atoms of the zeolite are represented in red and yellow, respectively.



SIMULATION METHODS Adsorption isotherms were computed by Monte Carlo simulations in the grand canonical ensemble (GCMC) in which chemical potential, volume, and temperature are fixed. Pressure is related through fugacity by the Peng−Robinson equation.13 The zeolites structures were considered to be rigid. The crystallographic positions for the different zeolites under study were taken from previous publications by van Koningsveld et al.14 for MFI, Fyfe et al.15 for MEL, Morris et al.16 for FER, and Marler17 for TON. The CHClFBr molecule was considered to be rigid as well and was modeled using the OPLS force field.18 The Coulombic interactions were computed by using Ewald summations with a relative accuracy of 10−6. The dispersive interactions of the adsorbate with the zeolite were modeled by Lennard-Jones (LJ) interactions between the atoms of the adsorbate and the oxygen atoms of the zeolite framework. LJ interactions were truncated and shifted at 12 Å. The polarity of the molecules and zeolites was modeled by partial charges on the atoms. Table 1 lists all LJ

Figure 2. Computed adsorption isotherms of the S enantiomers of CHClFBr as a pure component in MEL (circles), MFI (squares), FER (upward-pointing triangles), and TON zeolites (downward-pointing triangles) at 298 K.

Figure 3 shows the enantiospecific adsorption isotherms for chiral mixtures of CHClFBr in the same zeolite structures at the fixed values of pressure mentioned before. This implies that all zeolites were saturated and that the total number of CHClFBr molecules remained more or less constant during the runs, although the relative fractions of S and R enantiomers could

Table 1. Lennard-Jones Parameters and Partial Charges Used in This Worka atom

ε/kB (K)

σ (Å)

charge (e−)

Ozeo Oazeob Si Al Ca C H Cl F Br

48.202 48.202 − − 25.187 25.137 15.034 133.743 40.772 198.088

3.033 3.033 − − 3.093 3.600 2.500 3.470 2.750 3.650

−1.025 −1.200 2.050 1.750 2.000 0.307 0.102 −0.115 −0.189 −0.105

a

Interactions between nonidentical atoms are defined by the Lorentz− Bertheloz mixing rules. bOa corresponds to the oxygen atoms of the zeolites that are connected to aluminium atoms.

parameters and partial charges. In addition to the standard Monte Carlo moves, rotation, translation, regrow, reinsertion, and identity change, we also included ξ-replica exchange moves and chiral inversion moves,8 which considerably improve the statistics for the adsorption of chiral and multicomponent mixtures. All simulations were performed at 298 K and at constant pressures of 102 Pa for MEL, 10−2 Pa for TON, and 10 Pa for MFI and MEL, which correspond to the saturation pressures for the various structures.

Figure 3. Representation of the fractional adsorbed phase of SCHClFBr in the zeolites against the fractional content of the same component in the bulk phase. The solid line represents the results from MEL; dotted line, MFI; dashed line, FER; and double-dotted line, TON. 1525

dx.doi.org/10.1021/jp3111937 | J. Phys. Chem. C 2013, 117, 1524−1530

The Journal of Physical Chemistry C

Article

still fluctuate significantly. The data presented in Figure 3 show the fraction of S enantiomers in the adsorbed phase as function of the same fraction in the gas phase. Naturally, these curves must go through the points (0,0) and (1,1), as well as (1/2,1/2). The latter is a result of the symmetry of the system. If all atom coordinates were mirrored into a plane of symmetry of the zeolite, all of the chiral guest molecules would change identity. Because the energy of the system does not change as a result of this operation and the chemical potentials of the S and R enantiomers are identical for racemic mixtures, this mirror configuration must have exactly the same probability as its original. Hence, there cannot be an enantiomeric excess at this point. MEL, MFI, and FER show straight lines, indicating that the two components, R and S, are adsorbed at the same ratio as in the gas phase. This aselective adsorption corroborates previous results on apolar molecules showing that pure silica zeolites are not selective.9,10 However, differently from the apolar molecule studies, CHClFBr in TON does show a heteroselective deviation from the straight line. A rationalization of this difference is that the polar CHClFBr molecules have long-range interactions. For apolar molecules, the long-range interaction is established indirectly through the cations in the Al-substituted frameworks. Of course, having long-range interactions between enantiomers is not sufficient. These interactions should also be very specific, within the environment of the zeolite, to discriminate between R−R (or S−S) and S−R molecular interactions. Apparently, only TON is able to establish this distinction, in contrast to MEL, MFI, and even FER, which is very similar to TON. We also analyzed the effect of Al substitution. To quantify the enantioselective effect, we applied the adapted enantiomeric excess, ee*, as defined in ref 9, which graphically corresponds to the maximum vertical displacement, either positive or negative, from the diagonal in Figure 3. Homoselective adsorption results in a negative value, whereas heteroselective adsorption gives a positive value. Figure 4 shows that the Al substitution and concurrent inclusion of nonframework cations increases the selectivity in MFI, as in studies of apolar molecules. However, all ee* values are strictly positive (except for a small negative value for pure silica MFI), indicating that heteroselective adsorption is the only type of enantioselectivity for CHClFBr. For apolar chiral molecules, both homoselectivity and

heteroselectivity were observed. 9,10 To understand the mechanism of heteroselective adsorption with small polar molecules, we further focus on pure silica TON. We will try to rationalize why it is selective whereas the similar FER structure is aselective. For this purpose, we determined the carbon− carbon distance distribution for the different enantio pair combinations by averaging over all simulations, that is, for different S fractions in the gas phase. In Figure 5, one can distinguish three main distances between the carbon atoms, at around 4, 5.5, and 7 Å, which

Figure 5. Carbon−carbon distance between CHClFBr enantiomers. The solid (red), dashed (green), and dotted (blue) lines represent the distances of S−S, R−R, and S−R dimers, respectively.

correspond to the separation of cage centers in TON. Naturally, at large distances, there is no significant difference in the histograms of pairs having the same chirality, R−R/S−S versus the histogram for pairs of opposite chirality, S−R. However, at small distances, there is a noticeable difference. The S−R pair is able to get much closer than the R−R or S−S pair inside the TON zeolite. Related to this is the considerably higher occurrence of configurations with a C···C distance of less than 4 Å in S−R pairs as compared to R−R/S−S pairs, which is of sufficient magnitude to account for the observed heteroselective effect in Figure 3. To isolate the possible energetic origin of this effect, we analyzed minimum-energy dimer configurations in a vacuum using the Baker optimization scheme starting from several randomized initial conditions. As result, we obtained the global minima of the S−R and R−R (or S−S) dimers in addition to many local minima. Considering the global minima, we detected that the energy of the R−R dimer is lower than that of the S−R dimer (−1253.19 versus −1228.93 kBK). From these data alone, one would rather predict homoselective behavior instead of the observed heteroselectivity. However, based on a simple geometrical argument, one can exclude both global minimum configurations when considering the zeolite as well. Carbon−bromine is the longest bond in the CHClFBr molecule. One can consider the channels of TON as cylinders with a small radius of the same size. Then, it is evident that the dimer configurations, minimized in a vacuum, can fit in the zeolite channels only if the carbon−bromine bonds of the two molecules are almost parallel to the carbon−carbon intermolecular axis. This is not the case for the global minimum configurations, as one can see in Figure 6.

Figure 4. Enantiomeric excess values, ee*, for the zeolites of Figure 3 and all eight Al-substituted MFI systems using calcium as the extraframework cation. 1526

dx.doi.org/10.1021/jp3111937 | J. Phys. Chem. C 2013, 117, 1524−1530

The Journal of Physical Chemistry C

Article

shape of the channel having a cross section that is not completely cylindrical but rather elliptical. The cylindrical pore is narrower along the X axis, making its direction accessible only by the smallest bond. Based on these findings, we re-examined our set of minimized dimer configurations considering only the local minima having bromine as the outer atoms. If viewed from the top along the Br−C···C−Br axis, the two types of configurations can be characterized as what we call triangular and hexagonal. The first type has an R−R enantiomer as a minimum, whereas the second has S−R as the minimum (see Figure 8).

Figure 6. Snapshots of the optimized dimeric configurations R−R (top) and S−R (bottom) in a vacuum. Carbon atoms are represented in gray, hydrogen atoms in white, bromine atoms in red, chlorine atoms in green, and fluorine atoms in blue. Carbon−carbon distances are given in angstroms.

This hypothesis can be quantified by analyzing the angular distribution of the molecules inside the TON zeolite. We considered the angle between the C−Cl, C−H, C−F, and C− Br bonds and the Cartesian axes, with the Z axis parallel to the zeolite channel. As before, we averaged over all simulations. The distribution shown in Figure 7 confirms our geometrical argument. The cosine of the angle between the carbon− bromine bond and the Z axis is peaked around 1, indicating that the carbon−bromine bond is almost always parallel to the axis of channel. Interestingly, the angular distribution of the C−H bond with the X axis is even more peaked. This is due to the

Figure 8. Snapshots of the optimized dimeric configurations R−R (top) and S−R (bottom) in a vacuum with the Br−C···C−Br distribution. Carbon atoms are represented in gray, hydrogen atoms in white, bromine atoms in red, chlorine atoms in green, and fluorine atoms in blue. Distances between atoms are given in angstroms.

Figure 7. Angular orientation of CHClFBr in TON (α is the angle described by the C−X vector and the Cartesian axis). Solid (red), dashed (green), dash-dotted (blue), and dotted (pink) lines show the results for carbon−hydrogen, carbon−fluorine, carbon−chlorine, and carbon−bromine, respectively.

Figure 9 shows a cartoon representation of the top view of the R−R dimer that looks like a triangle, where the atoms are “superimposed” (their separation is mainly along the Z axis, which is perpendicular to the page). The two molecules show a small inclination that results in a larger separation distance between the two chlorine atoms with respect to the separation of the oppositely charged hydrogen and fluorine atoms. On the other hand, the S−R configuration viewed from the top looks hexagonal. By considering solely the Coulombic energy due to partial charges of neighboring atoms, one can decipher why these configurations are the lowest in energy for the triangular and hexagonal dimer configurations. If one considers all possible permutations of the atoms in the lower molecule with the atoms in the upper molecule held fixed, the energy always goes up. There are two negative charges and only one positive charge. For the triangular case, it is clear that, to obtain the lowest energy, one should superimpose the atoms with the positive charge on the most negatively charged atoms. This gives the highest possible negative Coulombic energy term twice and the smallest possible positive energy term (by the pairing of the smallest negatively charged atoms) once. 1527

dx.doi.org/10.1021/jp3111937 | J. Phys. Chem. C 2013, 117, 1524−1530

The Journal of Physical Chemistry C

Article

S−R dimer for the hexagonal structure. Again, the R−R triangular dimer energy is lower than the S−R hexagonal dimer, in contradiction to what one would expect because the adsorption is heteroselective. However, only the hexagonal structures were actually found in the MC simulations of CHClFBr molecules inside the zeolite. As expected, some of the structures were hexagonal R−R or S−S dimers, but the majority were S−R dimers. Hence, only hexagonal dimer configurations fit in the zeolite channel, and the lowest energy for these types of dimer configurations is obtained by the enantio pair S−R. These two facts together explain why S−R molecules can be much closer inside TON than R−R or S−S dimers and why the adsorption isotherm is heteroselective. Regarding this aspect, the only question remaining is: Why do the triangular dimer configurations not fit? To understand this situation, considering the zeolite channel as a cylinder with an elliptical cross-sectional area is insufficient. The smallest possible imaginary ellipse that encapsulates all atoms of Figure 9 is not smaller for the hexagonal structure than it is for the triangular structure. To understand the fact that no triangular configurations occur inside the zeolite, we performed the following analysis. We took a snapshot of one of the MC simulations in which a hexagonal S−R dimer configuration was detected (with a C···C distance of less than 4 Å). We eliminated all of the other molecules and minimized the structure. The minimization resulted in only small reorientations (e.g., C···C changed from 3.62 to 3.64 Å) and

Figure 9. Schematic representation of the top view of R−R (left) and S−R (right) dimeric configurations, with the R enantiomer on top. The partial charges (e−) of each atom are also indicated. Carbon atoms are represented in gray, bromine atoms in red, chlorine atoms in green, fluorine atoms in blue, and hydrogen atoms in white. Dashed bonds refer to the bonds of the lower molecule.

Similarly, for the hexagonal structure, one should position the positively charged atom of the lower molecule between the negatively charged atoms of the upper one. In addition, the most negatively charged atoms should be kept away from each other. This simple reasoning explains why the triangular configuration has an R−R (or S−S) dimer as the minimumenergy structure, whereas the minimum-energy structure is an

Figure 10. Top and side views of CHClFBr molecules in the optimized S−R (top) and initial R−R (bottom) dimeric configurations. Distances between atoms are given in angstroms. 1528

dx.doi.org/10.1021/jp3111937 | J. Phys. Chem. C 2013, 117, 1524−1530

The Journal of Physical Chemistry C

Article

a lower energy compared to the initial energy (−22644 versus −23514 kBK). Then, we took a triangular R−R dimer from the vacuum minimization and translated and rotated this configuration into an empty zeolite to obtain approximately the same orientation inside the zeolite as the hexagonal S−R dimer. That is, the C···C midpoint was matched, as well as the C···C angular orientation. Finally, a rotation along the C···C axis was performed to obtain the same angle between the C−H bond of the R enantiomer and the X axis. The geometry and energy of this artificially created zeolite−dimer configuration was then analyzed before and after energy minimization. The minimization caused a significant deviation from the initial conditions. The C···C distance increased from 3.8 to 4.8 Å, and the energy decreased from +1264 to −21061 kBK, indicating that the triangular configuration does not allow the two molecules to remain close when positioned inside the zeolite. This minimized structure was therefore disregarded in the further analysis. Figure 10 shows the top and side views of the minimized hexagonal S−R configuration inside the channel of TON and of the initial triangular R−R configuration. Regarding the top views, it is far from obvious why the triangular configuration does not fit. None of the atoms in the triangular configuration seem to be much closer to the “cylinder wall” than the atoms in the hexagonal configuration. Only if one inspects the side views of the configurations inside the channel can a major mismatch be detected. The initial value of one of the chlorine−oxygen distances is 2.38 Å for the triangular R−R configuration, which is in the repulsive region of their LJ interactions. In the top view, all chlorine atoms seem to be equally close to the oxygens of the zeolite, but this is deceptive because the relative orientations along the Z direction are not visible. The hexagonal structure with C−Cl bonds pointing in the positive and negative Y directions avoids Cl−O distances that are too small by having the two Cl atoms at different Z values with respect to the closest oxygen inside the zeolite ring. The triangular configuration, with both chlorine atoms pointing in the positive Y direction, is not able to do the same. One of the Cl atoms will meet a nearby oxygen inside the cylindrical wall having approximately the same positioning along the Z axis, which causes a high penalty to the total energy of the system. Hence, the cylindrical model is not sufficient to describe the observed enantioselectivity. Both the elliptical shape of the cross-sectional area and the corrugation of the potential energy field along the channel are important parameters to understand the heteroselectivity of this system. Finally, we briefly discuss why enantioselectivity is absent in FER even though it has a similar channel-type structure with even narrower pores. Naively, one would expect that the smaller channel diameter would enhance the subtle packing differences between R−R (or S−S) and S−R dimers. However, although the top view indeed suggests that FER has narrower pores than TON (see Figure 11), the side view shows more open windows at the sides of the principal channels in the X−Z plane. This particular topology gives too much rotational freedom to the CHClFBr molecules, thereby destroying any possible enantiospecific packing effect.

Figure 11. Comparison of TON (top) and FER (bottom) zeolites, including views of the X−Y (left), Y−Z (center), and X−Z (right) planes.

power, but it is difficult to exploit this effect for industrial applications. Doing so requires a special siting of aluminum atoms in the framework that is not yet under control in the synthesis processes of zeolites. Moreover, chiral alkanes presumably have very low diffusion constants in MFI zeolites. In contrast, the small chiral polar molecules studied in this work do not depend on Al substitutions to show enantioselectivity and diffuse much more rapidly. This brings technological applications closer to reality. The effect of selectivity has a strong dependence on topology, because zeolites that are assumed to be similar can show different adsorption behaviors. For instance, both TON and FER are zeolites containing nonintersecting channels, but only TON is selective for the studied molecule. The molecular mechanism responsible for heteroselectivity in TON is fundamentally different from that of chiral apolar molecules in Al-substituted zeolites. The reason for the difference is that polar molecules have long-range interactions and, therefore, do not need to rely on indirect interactions through nonframework cations. The mechanism is explained by the fact that enantiomers of opposite chirality are able to approach each other at much smaller distances than enantiomers of identical chirality within the TON zeolite channel. The standard representation of the channel of the zeolite as a simple cylinder is not sufficient to explain this effect. The ellipticity of the cross section and corrugation of the potential field along the channel’s axis are additional parameters that are needed to describe this enantioselective packing effect. This understanding allows for the rationalization of the best material and conditions for the optimal enantiopurification and separation of chiral mixtures.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: +47 735 94142.



Notes

The authors declare no competing financial interest.

CONCLUSIONS In this work, we have shown the enantioselective capacities of achiral zeolites when adsorbing small polar molecules. The previous studies with nonpolar molecules were the first proof of principle that achiral zeolites can have chiral discrimination



ACKNOWLEDGMENTS This work was supported by the Spanish ”Ministerio de Ciencia e Innovacion” (CTQ2010-16077/BQU), the “Junta de 1529

dx.doi.org/10.1021/jp3111937 | J. Phys. Chem. C 2013, 117, 1524−1530

The Journal of Physical Chemistry C

Article

Andalucia” (P07-FQM-02595), and the European Research Council through an ERC Starting Grant (Project ERCStG’11RASPA). A.M.-C. thanks the Spanish “Ministerio de Educacion” for her predoctoral fellowship and the European Science Foundation, SimBioMa project, for Exchange Grants. T.S.v.E. acknowledges support from the Research Council of Norway.



REFERENCES

(1) Maier, N. M.; Franco, P.; Lindner, W. J. Chromatogr. A 2001, 906, 3−33. (2) Gübitz, G.; Schmid, M. G. J. Chromatogr. A 2008, 1204, 140− 156. (3) Breuer, M.; Ditrich, K.; Habicher, T.; Hauer, B.; Keßeler, M.; Stürmer, R.; Zelinski, T. Angew. Chem., Int. Ed. 2004, 43, 788−824. (4) Blaser, H.-U.; Müller, M. Stud. Surf. Sci. Catal. 1991, 59, 73−92. (5) Li, Y.; Yu, J.; Wang, Z.; Zhang, J.; Guo, M.; Xu, R. Chem. Mater. 2005, 17, 4399−4405. (6) Tang, L.; Shi, L.; Bonneau, C.; Sun, J.; Yue, H.; Ojuva, A.; Lee, B.L.; Kritikos, M.; Bell, R. G.; Bacsik, Z.; Mink, J.; Zou, X. Nat. Mater. 2008, 7, 381−385. (7) Sun, J.; Bonneau, C.; Cantin, A.; Corma, A.; Diaz-Cabanas, M. J.; Moliner, M.; Zhang, D.; Li, M.; Zou, X. Nature 2009, 458, 1154−U90. (8) van Erp, T. S.; Dubbeldam, D.; Caremans, T. P.; Calero, S.; Martens, J. A. J. Phys. Chem. Lett. 2010, 1, 2154−2158. (9) Caremans, T. P.; van Erp, T. S.; Dubbeldam, D.; Castillo, J. M.; Martens, J. A.; Calero, S. Chem. Mater. 2010, 22, 4591−4601. (10) van Erp, T. S.; Caremans, T. P.; Dubbeldam, D.; Martin-Calvo, A.; Calero, S.; Martens, J. A. Angew. Chem. 2010, 122, 3074−3077. (11) van Erp, T. S.; Dubbeldam, D.; Calero, S.; Martens, J. A. Chem. Eng. Sci. 2010, 65, 6478−6485. (12) Castillo, J. M.; Vlugt, T. J. H.; Dubbeldam, D.; Hamad, S.; Calero, S. J. Phys. Chem. C 2010, 114, 22207−22213. (13) Robinson, D. B.; Peng, D.-Y.; Chung, S. Y.-K. Fluid Phase Equilib. 1985, 24, 25−41. (14) van Koningsveld, H.; van Bekkum, H.; Jansen, J. C. Acta Crystallogr. B 1987, 43, 127−132. (15) Fyfe, C. A.; Gies, H.; Kokotailo, G. T.; Pasztor, C.; Strobl, H.; Cox, D. E. J. Am. Chem. Soc. 1989, 111, 2470−2474. (16) Morris, R. E.; Weigel, S. J.; Henson, N. J.; Bull, L. M.; Janicke, M. T.; Chmelka, B. F.; Cheetham, A. K. J. Am. Chem. Soc. 1994, 116, 11849−11855. (17) Marler, B. Zeolites 1987, 7, 393−397. (18) Costante-Crassous, J.; Marrone, T. J.; Briggs, J. M.; McCammon, J. A.; Collet, A. J. Am. Chem. Soc. 1997, 119, 3818−3823.

1530

dx.doi.org/10.1021/jp3111937 | J. Phys. Chem. C 2013, 117, 1524−1530